新型AI工具可生成患者的“数字孪生体”预测健康状况

Core Insights - A research team from the University of Melbourne has developed an AI tool called "Digital Twin-Generative Pre-trained Transformer" (DT-GPT) that creates digital twins of patients and successfully predicts individual health conditions [1][2] Group 1: AI Tool Development - The DT-GPT model was trained using three datasets covering thousands of electronic health records [1] - The model analyzes medical data from patients with Alzheimer's disease, non-small cell lung cancer, and those in intensive care [1] Group 2: Predictive Capabilities - DT-GPT can generate a digital twin of patients and analyze their medical history, lab results, diagnoses, and treatment plans [1] - The model demonstrated superior accuracy in predicting health status changes compared to 14 other advanced machine learning models [1] - For over 35,000 ICU patients, DT-GPT accurately predicted changes in magnesium levels, blood oxygen saturation, and respiratory rates within 24 hours based on previous lab results [1] Group 3: Clinical Implications - This AI tool enables doctors to anticipate patient health deterioration, allowing for early intervention [2] - It can predict drug side effects and assist in customizing treatment plans based on individual patient characteristics, improving treatment outcomes [2] - The technology paves the way for a shift from passive to predictive and personalized medicine, potentially disrupting the clinical trial landscape [2]